Sarah Catanzaro (@sarahcat21) started following @GordonJo76 on Sep 13, 2024

RT @lukaszkaiser: I'm so happy to see o1 launch! Leading this research with my colleagues for almost 3 years and working on related ideas e…

RT @karlcobbe: Very excited to share our latest work @OpenAI teaching models how to solve hard reasoning tasks! https://t.co/LNzdEylIUM

RT @MillionInt: Let the great ones cook

this, but unironically

☺️

Check out the clip for a preview of behind-the-scene stories of 🍓

Such an honor to have an opportunity to work together and learn from these researchers! This video doesn't show all the great people who worked on this project. Please check out https://t.co/rRkUhjAiFT

RT @sama: @AiBreakfast no

More seriously, I think the most interesting questions in AI right now include how to allocate compute resources/budget across pretraining, post-training, & inference. Or alternatively, how to reduce the cost of experimentation/iteration so we can make these decisions better.

Damnit- apparently yesterday was the day to tweet my thoughts on test time compute so I could seem smarter than gpt-o1

RT @devahaz: Taylor Swift: I’m endorsing Harris in part because I believe a woman’s right to choose is important Elon: I will impregnate y…

RT @AlecStapp: The pre-industrial era https://t.co/opGXHPKdPC

If you think Biden bowed out gracefully then read this incredibly entitled and firmly detached from reality interview with Anita Dunn. https://t.co/9iH5elK32R

Biden’s not running because he was gonna lose, the debate made that obvious, and fortunately for Democrats they have adults in the room and Pelosi et. al. had the smarts and guts to play hardball and use as much leverage as they could to force him out. https://t.co/7wjo6MU5xH

Dear Amplify founders, please don’t tell me you have concepts of a plan at our next board meeting 🤦🏻‍♀️

RT @TristanThrush: Do you want to select great LLM pretraining data but don’t have 1000 H100s for a ton of mixture experiments? What about…

RT @corbtt: Spoke to a Microsoft engineer on the GPT-6 training cluster project. He kvetched about the pain they're having provisioning inf…

I’m very proud of what @richardartoul, @ryanworl & the @warpstream_labs team did in such a short period of time, grateful they chose us as their first investor, & excited by their ambitious roadmap. @jaykreps you got some great ones; Confluent will kill it with them on board.

This is a team that makes a deep impact fast - and with additional resources, I have no doubt they will defy imagination - continuing to challenge our assumptions about how we interface with/leverage the cloud and deliver tools developers love.

Last March, Richie & Ryan were contemplating starting a company but weren’t sure if they were ready to take the plunge. In just over a year, they’ve hired an amazing team, built a product that changes how developers use stream processing, and engaged sophisticated customers.

RT @bryce: Of course Virgil was up on @jugendingenieur before everyone https://t.co/TpeIMRciac

RT @opiumhum: The new @jugendingenieur mic is an absolute piece of art and probably the most innovative and interesting microphone to be re…

RT @anabrandusescu: "Only 65 randomized controlled trials of AI interventions were published between 2020 and 2022, a review shows. Meanwhi…

RT @PleaseBeGneiss: this is a very strong case where shitposting on your phone is not a waste of time and can literally shape the future of…